Managing managers works when you stop directing the work and start developing the person doing the directing. Your first GPM role means your decisions now reach customers through someone else's judgment, not yours — so your job shifts from being right to making your GPM capable of being right, most of the time, without you.
Manage your GPM's decision-making, not their team's roadmap. Skip-level only for safety, ethics, or a genuinely stuck escalation — otherwise coach, don't correct. Your scorecard is now the health of teams you don't touch directly.
The Second-Order Leap: Why Managing Managers Feels Like Starting Over
Managing managers is a second-order job: you no longer touch the product directly, you influence it through a person who influences it through their own team. Ram Charan's Leadership Pipeline calls this Passage Three — the leap where a manager's success starts depending on skills they've rarely had to practice: coaching, delegation, and organizational judgment.
If you're newer to the director seat in general, our complete guide to the director of product role covers the broader shift; here we isolate the one piece that trips up almost everyone — the leap from managing doers to managing managers.
Every promotion up to this point rewarded the same instinct: get closer to the problem, make the call yourself, ship it. Becoming a director of directors punishes that instinct. Andy Grove's High Output Management frames a manager's output as the output of the organization they influence — for a director managing GPMs, that organization now sits two teams removed from your own hands.
Three things change at once when you take on your first GPM:
- Your unit of work becomes a person's judgment, not a roadmap decision.
- Your feedback loop lengthens — a bad call by a GPM's team might not surface for a full quarter.
- Your visibility drops — you see summaries and escalations, not the daily texture of the work.
Camille Fournier's The Manager's Path describes this exact discomfort: the vertigo of realizing your reports' reports are now, functionally, your responsibility too, without you ever having met most of them.
The Core Rule: Manage the Manager, Not Their Team
The core rule is simple to state and hard to hold: your job is developing your GPM's judgment, not running their team's roadmap, rituals, or reviews yourself. Every time you make a call that belongs to your GPM, you don't just fix one problem — you teach their team that the real decision-maker sits one level up.
This is fundamentally a boundary question — where does your remit end and your GPM's begin? Our team boundaries guide maps this at the org-chart level; inside a single reporting line, the same logic governs meetings, decisions, and even casual Slack habits.
The clearest way to hold the boundary is to write down, once, who owns which category of decision — because in the moment, urgency makes almost everything feel like your call to make.
| Decision or Activity | Owned By | Your Role |
|---|---|---|
| Team's sprint prioritization | GPM | Ask about tradeoffs in 1:1s; don't dictate them |
| Hiring or exiting an IC PM | GPM, with your sign-off | Coach the decision; approve the outcome |
| The GPM's own performance and growth | You | Direct coaching, feedback, career planning |
| Cross-team roadmap conflicts | You | Arbitrate — GPMs can't out-rank each other |
| Feedback to an IC PM on their work | GPM | Route it through the GPM; don't deliver it yourself |
| Org design changes (splitting or merging teams) | You | Owned at your level, communicated jointly |
Notice the pattern: nearly everything inside a single team's execution belongs to the GPM. Only decisions that cross teams, or that concern the GPM as a person, are actually yours.
The First 90 Days With a New GPM
A brand-new GPM — whether promoted from your own team or hired in — needs more explicit boundary-setting than an established one, precisely because the old habit of solving things directly is freshest for both of you. Say the rule out loud in your first 1:1: "I'll ask you about your team's decisions, but I won't make them."
That single sentence, repeated a few times in the first quarter, does more to set the boundary than any org chart. It also gives the GPM permission to push back the first time you slip and start directing instead of coaching.
A short written charter helps too — a half-page note covering what decisions the GPM owns outright, what needs a heads-up but not approval, and what genuinely requires your sign-off. It sounds bureaucratic for a relationship built on trust, but a first-time GPM under pressure will default to over-checking with you unless the boundary is explicit somewhere they can reread it.
The Skip-Level Framework: When to Step In vs. When to Stay Out
Skip-level when there's a safety, ethics, or legal issue; when a GPM is genuinely stuck after real effort and asks for help; or when a cross-team conflict needs a tie-breaker only you can provide. Stay out when a decision is reversible, contained to one team, and simply not the one you'd have made yourself.
A short framework beats a gut call made under time pressure, especially the first few times a GPM's team hits a wall.
| Situation | Default Move | Why |
|---|---|---|
| An IC PM emails you directly to complain about their GPM | Redirect to the GPM first, then check in with the GPM separately | Bypassing trains the team to route around its own manager |
| A GPM's team is about to ship something that will visibly hurt a customer this week | Step in directly, then debrief with the GPM afterward | Some risks are too costly to use as a teaching moment |
| A GPM asks, "What would you do here?" on a reversible call | Ask what they've already considered, then let them decide | Answering trains dependency; questioning builds judgment |
| Two GPMs disagree on a shared roadmap dependency | Skip-level a joint conversation and arbitrate | Peers can't out-rank each other; only you can break the tie |
| A GPM seems stressed or disengaged for weeks | Skip-level a 1:1 focused on them, not their team's metrics | People risk needs direct attention, not proxy metrics |
| A GPM's team keeps missing the same kind of deadline | Coach the GPM on the pattern; don't manage the team's calendar | The GPM owns execution; you own their diagnostic skill |
Only two of these six situations call for you to act directly on the team. Everything else routes back through the GPM, even when it would be faster to just do it yourself.
Build the safety-valve version of this into your regular rhythm instead of inventing it under pressure. A standing skip-level 1:1 with a senior IC on each team, scheduled on the cadence our operating cadence and rhythms guide describes, gives you a read on team health without forcing you to insert yourself reactively. And if you trust the bar that got this GPM promoted — the judgment-first bar covered in interviewing for judgment — most of these situations resolve themselves before they ever reach you.
A Simple Test Before You Act
Before stepping into a GPM's team, ask three questions in order: Is anyone at risk right now? Is this reversible within a normal cycle? Has the GPM actually tried to solve it first? A "yes, no, yes" answer almost always means stay out; a "yes" to the first question overrides everything else.
The Trap: Solving Your GPM's Problem for Them
The most common failure mode for new directors is answering a GPM's question instead of teaching them how to find the answer themselves. It feels helpful and efficient in the moment; over months, it quietly signals to the GPM's own team that decisions really get made one level up, which erodes the GPM's authority every time it happens.
Picture the pattern: a GPM comes to you with a prioritization conflict between two of their PMs. You have the context, the opinion, and the authority, so you just decide. The conflict resolves in ten minutes — but the two PMs now know their GPM wasn't the one who actually settled it.
L. David Marquet's Turn the Ship Around describes the antidote as intent-based leadership: pushing the authority to decide down to whoever has the competence and context to decide well, and building that competence deliberately if it's missing yet. Solving it yourself does the opposite — it centralizes authority exactly where competence is supposed to grow.
If a GPM's team keeps chasing the wrong problem, don't run their next discovery sprint for them either. Teach them to diagnose it — using something like the opportunity-scoring approach from our jobs-to-be-done complete guide or the emotion-curve mapping in our customer journey complete guide — so the next miss is one the GPM catches before you do.
Watch for these tells that you've slipped into solving instead of coaching:
- You remember the details of a GPM's team roadmap better than the GPM does.
- ICs cc you on emails "just in case," instead of trusting their GPM to escalate.
- Your 1:1s with a GPM are mostly you giving answers, rarely you asking questions.
- A GPM stops bringing you unsolved problems, because last time you just took over.
Each of these is reversible. The fix is almost always the same: ask one more question before you answer one. "What have you already tried?" and "What would you do if I weren't in the room?" cost you thirty seconds and hand the decision back.
The Quiet Cost Nobody Puts on a Scorecard
The damage from solving a GPM's problem rarely shows up the week it happens. It shows up two quarters later, when the GPM still can't run a hard conversation alone, and you realize you never actually let them practice on the easy ones.
The Mental Shift: Your Success Now Lives Two Layers Down
Your scorecard as a director of GPMs is no longer your own decisions — it's the decision quality of people your GPMs manage, most of whom you rarely speak to directly. Gallup's long-running research on manager impact puts a manager's influence on team engagement at roughly 70 percent of the variance, and that influence now runs through two layers, not one.
This is disorienting because you can't inspect it directly. You won't sit in on most of the 1:1s your GPM has with their ICs. You won't read most of the specs their team ships. The only lever left in your hands is the quality of the person standing between you and that work.
Track these instead of the work itself:
- Retention and internal mobility inside each GPM's team — a GPM who develops people keeps them or promotes them; one who doesn't loses them quietly.
- How often escalations skip the GPM — a rising
escalation rateis an early signal of eroding trust in that layer. - The GPM's own decision quality on calls you never touched — ask them to walk you through one after the fact, not before.
- Whether the GPM is coaching or directing their own ICs — the same second-order pattern repeats one level further down.
Marshall Goldsmith's What Got You Here Won't Get You There makes the point almost word for word: the behaviors that got you promoted to director — fast, confident, hands-on calls — are frequently the exact behaviors that now cap how far your GPMs can grow.
This feels unfair at first, and it's worth naming that plainly rather than pretending it doesn't. You're being evaluated on outcomes produced by decisions you never saw, made by people two conversations removed from you. The adjustment isn't liking that arrangement — it's building enough trust in your GPMs' judgment that you stop needing to see the decision to trust the outcome.
Recognizing Your Own Default Management Mode Before You Flex It
Most directors default to one management mode under pressure — directive, coaching, or hands-off — no matter what a situation actually calls for. The real skill is naming your default, then flexing away from it: directive with a brand-new GPM, coaching with a capable but conflict-avoidant one, hands-off with a GPM who has already earned the room.
The three modes look different in practice:
- Directive — you set the direction and check the work closely. Right for a brand-new GPM, a genuine crisis, or a decision with real safety stakes.
- Coaching — you ask questions, offer frameworks, and let the GPM make the call. The default mode for most healthy, ongoing GPM relationships.
- Hands-off — you stay informed but don't intervene unless asked. Right for a GPM who has already earned the room through a track record of good calls.
Key Takeaways
- Manage the manager, not their team — nearly every decision inside a GPM's team belongs to the GPM; your job is their judgment, not their roadmap.
- Skip-level for safety, ethics, or a genuine stuck point — not because you'd have made a different, equally reversible call.
- Solving a GPM's problem for them is the most common trap — it feels efficient once and costs their authority every time after.
- Your scorecard is now two layers down — track retention, escalation patterns, and GPM decision quality instead of inspecting the work directly.
- The behaviors that got you promoted often cap your GPMs' growth — fast, confident, hands-on calls need to become questions, not answers.
- Know your own default management mode before you try to flex it — most directors default to taking over under pressure without noticing it.
Frequently Asked Questions
How do you know when to skip-level as a director managing GPMs?
Skip-level when there's a safety, ethics, or legal risk; when a GPM has genuinely tried and is stuck; or when two GPMs need a tie-breaker they can't provide each other. If the call is reversible, contained to one team, and simply not the choice you'd have made, stay out and let it play out.
What's the actual difference between managing an IC PM and managing a GPM?
Managing an IC PM is about their product decisions; managing a GPM is about their people decisions. A GPM's job is coaching PMs the way you coach them, so your feedback has to focus on their coaching skill, delegation habits, and judgment — not just their roadmap calls.
How many GPMs can one director manage well?
Most directors hold this steady around four to six direct GPM reports, in line with the span-of-control ranges Elad Gil's High Growth Handbook and general management research both suggest for manager-of-managers roles. Fewer if teams are new or in crisis; more once GPMs are established and need less coaching.
Is it normal to miss doing hands-on product work after becoming a director?
Yes — nearly every new manager-of-managers goes through this, and it's a well-documented transition cost in Ram Charan's Leadership Pipeline model, not a sign you're in the wrong role. The discomfort usually fades once you find satisfaction in a GPM's growth rather than in the work itself.
How do you break the habit of solving problems for your GPM?
Ask one more question before you give one more answer — "What have you already tried?" or "What would you do if I weren't here?" hands the decision back in seconds. Do this consistently for a few weeks, and a GPM who's been leaning on you starts bringing fewer pre-solved requests and more real judgment calls.